2. I
t is well known that physical proper-
ties of oxide and semiconductor films
depend on the crystalline perfection.
Single crystalline, textured, and polycrys-
talline films of the same composition have
different resistivity, index of refraction, and
even band gap energy. Achieving the film
with the desired properties requires a num-
ber of calibration runs followed by the ex-
tensive ex-situ characterization. First,
crystalline structure of the films deposited
by evaporation, sputtering, chemical vapor
and pulsed laser deposition (CVD and
PLD) is assessed using X-ray diffraction,
scanning and transmission electron mi-
croscopy (SEM and TEM). Next, deposi-
tion parameters are adjusted according to
the characterization results. However, this
approach is extremely time consuming and
often suffers from the low run-to-run re-
producibility. Clearly, a technique that can
be used to characterize films in-situ during
the deposition will simplify this process
significantly.
Such a technique – Reflective High En-
ergy Electron Diffraction (RHEED) – was
introduced 80 years ago [1]. However, it
became widely used only in the early 80’s
with the development of one of the most
advanced thin film growth techniques –
Molecular Beam Epitaxy (MBE) [2]. The
schematic of a RHEED system installed in
the MBE chamber is shown in Figure 1a.
It consists of a high-energy electron gun
(10 – 30keV) directed towards the sub-
strate surface under a small incident angle
(1 - 3°). The well-collimated electron beam
is diffracted by the crystalline surface and
is detected by the fluorescence on the
phosphorus screen. The glancing incident
angle ensures that electrons penetrate only
through the top few monolayers making
this technique extremely surface sensitive.
In addition, glancing angle keeps both
electron gun and detector (phosphorous
screen) well clear of the material sources,
such as effusion cells in MBE and targets
in PLD. Thus, crystallographic structure of
the film can be studied during the deposi-
tion process.
RHEED pattern depends on the sur-
face morphology, reconstruction, and
crystal orientation. In general, four types
of pattern are collected from the crys-
talline surface.
59
3. Figure 2. RHEED diffraction patterns for a) single crystalline film with
the atomically flat surface consisting of the micrometer-size terraces,
b) single crystalline film with the atomically flat surface consisting of
the wide density of small terraces, c) single crystalline film with rough
three-dimensional surface, d) polycrystalline or textured film.
form into the diffraction streaks with the streak width being
inversely proportional to the terrace size, as shown in Figure 2b.
Single crystalline films with rough three-dimensional surface
Figure 1. a) Schematic representation of a RHEED system installed in produce transmission pattern – the electron beam now penetrates
the MBE chamber. b) Schematic representation of a high-pressure through the crystallites - a set of broad spots, as shown in Figure
RHEED system installed in the PLD chamber. 2c. This pattern is usually observed during the growth
Single crystalline films with the atomically flat surface con- of the strongly lattice mismatched materials – when the film
sisting of the micrometer-size terraces produce a set of sharp dif- grows via island (Volmer-Weber) or layer plus island (Stranski-
fraction spots lying along the Laue rings. Only the bottom half of Krastanov) mode. A transition from the diffraction streaks to the
the zero-order ring is usually visible due to the limited size of the transmission spots indicates surface roughening while transition
screen and the shadow effect from the substrate, as shown in Fig- from the transmission to the diffraction pattern is the sign of the
ure 2a. The geometry of the RHEED pattern can be simulated improvement of surface morphology through the coalescence of
using the kinematical theory. Then, the separation between the neighboring islands.
diffraction spots (r) can be related with the reciprocal lattice (a*) Polycrystalline films and textured films with no in-plane
or in-plane lattice (aII) spacing of the film through the equations: orientation produce a set of concentric circles, as shown in Fig-
r = a*λL/2π
ure 2d. If some in-plane orientation is present, as in the case of a
r = λL/aII
biaxially textured film, circles break into arcs. In addition, the pat-
tern can consist of a superposition of streaks and spots,
where λ is the electron energy and L is the distance between as shown for the MgO film in Figure 3b. Here, diffraction streaks
the substrate and the detector. Thus, it is possible to measure arise from the epitaxial MgO film and circles are from the
the in-plane lattice constant and, by monitoring RHEED pattern randomly oriented MgO inclusions, as was determined from the
evolution, study strain relaxation during the film growth and cross-sectional TEM image shown in Figure 3c.
determine its critical thickness [3,4]. Since the substrate can RHEED is also routinely used for the growth rate calibration.
be rotated around the z-axis, information can be collected along If the film grows in a layer-by-layer (Frank-van-der-Merwe)
different crystal directions. mode, intensity of the diffracted beam oscillates with the fre-
Due to the crystal symmetry interruption, topmost atomic layer quency proportional to the growth rate. The mechanism of the
of the film rearranges to minimize free surface energy. This sur- RHEED intensity oscillations is illustrated in Figure 4 [5].
face rearrangement, known as reconstruction, is also often visible Intensity of the beam diffracted by the extremely flat substrate is
in the RHEED as a weak superstructure. For example, a three- maximal prior to the growth. As the growth is initiated, scattering
fold reconstruction - 3 weak lines between major diffraction spots from the small two-dimensional islands nucleating on the surface
- that is due to the ordered oxygen vacancies on the (100) TiO2 sur- decreases the diffracted beam intensity. It approaches minimum
face is shown in Figure 3a. Since surface reconstructions are tem- at a half-monolayer coverage that is the roughest stage of the
perature-sensitive they, in addition to the thermocouple and growth. As the first monolayer is complete, the surface flattens
pyrometer readings, are used to monitor real surface temperature again by the coalescence of the islands and diffracted beam in-
and optimize growth conditions. tensity recovers. However, since the second monolayer always
When the terraces decrease in size, the diffraction spots trans- nucleates before the first is complete, the growth front roughens.
60 vtcmag@optonline.net August 2008 • Vacuum Technology & Coating
4. Figure 4. Mechanism of RHEED intensity oscillations during the layer-
by-layer growth [5].
a second differential pumping stage is added between the gun and
Figure 3. a) RHEED pattern collected from the (100) TiO2 showing
the chamber. A pumped tube protrudes inside the vacuum cham-
a three-fold reconstruction, b) RHEED pattern collected from the ber. The differential pumping aperture at the end of the tube is
epitaxial MgO film containing randomly-oriented inclusions inclusions, located close to the substrate. This decreases distance electrons
c) Cross-sectional TEM image of the same MgO film. travel through the high-pressure region, allowing efficient
RHEED operation even at high pressures [9].
Thus, RHEED oscillations slowly decrease in intensity during the
Extended to the high-pressure region, RHEED becomes an
growth, and eventually disappear. Thus, while oscillation fre-
invaluable tool for the in-situ monitoring during PLD, CVD, and
quency is the direct measure of the growth rate, the decay rate
sputtering. High quality differentially-pumped RHEED systems
provides rich information on the surface mobility of arriving
are now available from wide range of companies including Staib
atoms and can be used to study how growth temperature, over-
Instruments, Pascal Technologies, and Twente Solid State Tech-
pressure, and / or surfactants influence film growth [6].
nology. In addition, a number of excellent books discussing var-
Although it is assumed that electrons interact elastically with
ious RHEED aspects in details [10,11] and programs for
the crystal lattice, some of them undergo inelastic scattering and
automatic data acquisition and analysis [12] make this technique
loose energy. By adding grids, phosphorus screen can modified
extremely user-friendly.
into an electron energy loss detector, allowing combined RHEED-
electron energy loss spectroscopy (EELS) measurements [7].
References
Thus, surface morphology and chemistry can be simultaneously
analyzed. In addition, intensive cathodoluminescence (CL) is ex- 1. S. Nishikawa and S. Kikuchi, Nature 122, 726 (1928).
cited when the direct band gap semiconductor (AlxGa1-xN, ZnxCd1- 2. M.A. Herman and H. Sitter, Molecular Beam Epitaxy, Springer,
xSe) and oxide (ZnxMg1-xO) alloys are deposited. CL spectra can Berlin (1996).
be collected either in the imaging mode, via CCD camera 3. Y. Hida, T. Tamagawa, H. Ueba, C. Tatsuyama, J. Appl. Phys. 67,
equipped with special filters, or in the spectral mode with the lu- 7274 (1990).
minescence being resolved through the spectrometer and detected 4. H. J. Osten and J. Klatt, Appl. Phys. Lett. 65, 630 (1994).
by the photomultiplier tube [8]. CL provides additional informa- 5. K. Ploog, Angewandte Chemie - Int. English Ed. 27, 593 (1988).
tion on the alloy composition, uniformity, and film quality. 6. J.M. Van Hove, P. R. Pukite, P. I. Cohen, J. Vac. Sci. Technol. B 3, 563
RHEED was originally used only during the high vacuum dep- (1985).
osition processes (< 105 Torr), such as MBE. Two factors limited 7. W. Braun, L. Däweritz, K. H. Ploog, Physica E 2, 878 (1998).
its application at higher pressures. First, a high gas pressure (10-5 8. K. Lee, E. D. Schires, T. H. Myers, Mater. Res. Soc. Symp. Proc. 892,
– 10-3 Torr) strongly reduced filament lifetime. At even higher 0892-FF04-01.1 (2006).
pressures (∼ 10-2 Torr), increased scattering of electron beam by 9. J. Klein, C. Höfener, L. Alff, R Gross, Supercond. Sci. Technol. 12,
the gas molecules produced diffuse pattern. However, this limita- 1023 (1999).
tion was recently overcome through the development of a high- 10. W. Braun, Applied RHEED, Springer, Berlin (1999).
pressure system. At pressures below 10 mTorr, such a system 11. A. Ichimiya and P. I. Cohen, Reflective High Energy Electron
includes a single differential pump next to the electron gun. Due Diffraction, Cambridge University Press, Cambridge (2004).
to the high differential pumping efficiency, the pressure in the
electron gun is kept at ∼ 10-5 Torr level even when a chamber pres-
12. For example, kSa 400 by K-Space Associates, Inc., SAFIRE by
CreaTec Fischer & Co., and EZ-RHEED by MBE Control Solutions.
sure is in the 10-3 Torr range. A more advanced system, shown in
Figure 2b, is used at higher pressures (10 mTorr -100 mTorr)
when scattering of electron beam inside the chamber becomes
important. The RHEED screen is shifted closer to the sample and
Vacuum Technology & Coating • August 2008 www.vactechmag.com or www.vtcmag.com 61